Home | History | Annotate | Line # | Download | only in netinet6
ip6_input.c revision 1.217
      1 /*	$NetBSD: ip6_input.c,v 1.217 2020/06/19 16:08:06 maxv Exp $	*/
      2 /*	$KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $	*/
      3 
      4 /*
      5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
      6  * All rights reserved.
      7  *
      8  * Redistribution and use in source and binary forms, with or without
      9  * modification, are permitted provided that the following conditions
     10  * are met:
     11  * 1. Redistributions of source code must retain the above copyright
     12  *    notice, this list of conditions and the following disclaimer.
     13  * 2. Redistributions in binary form must reproduce the above copyright
     14  *    notice, this list of conditions and the following disclaimer in the
     15  *    documentation and/or other materials provided with the distribution.
     16  * 3. Neither the name of the project nor the names of its contributors
     17  *    may be used to endorse or promote products derived from this software
     18  *    without specific prior written permission.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     30  * SUCH DAMAGE.
     31  */
     32 
     33 /*
     34  * Copyright (c) 1982, 1986, 1988, 1993
     35  *	The Regents of the University of California.  All rights reserved.
     36  *
     37  * Redistribution and use in source and binary forms, with or without
     38  * modification, are permitted provided that the following conditions
     39  * are met:
     40  * 1. Redistributions of source code must retain the above copyright
     41  *    notice, this list of conditions and the following disclaimer.
     42  * 2. Redistributions in binary form must reproduce the above copyright
     43  *    notice, this list of conditions and the following disclaimer in the
     44  *    documentation and/or other materials provided with the distribution.
     45  * 3. Neither the name of the University nor the names of its contributors
     46  *    may be used to endorse or promote products derived from this software
     47  *    without specific prior written permission.
     48  *
     49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     59  * SUCH DAMAGE.
     60  *
     61  *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
     62  */
     63 
     64 #include <sys/cdefs.h>
     65 __KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.217 2020/06/19 16:08:06 maxv Exp $");
     66 
     67 #ifdef _KERNEL_OPT
     68 #include "opt_gateway.h"
     69 #include "opt_inet.h"
     70 #include "opt_inet6.h"
     71 #include "opt_ipsec.h"
     72 #include "opt_net_mpsafe.h"
     73 #endif
     74 
     75 #include <sys/param.h>
     76 #include <sys/systm.h>
     77 #include <sys/mbuf.h>
     78 #include <sys/domain.h>
     79 #include <sys/protosw.h>
     80 #include <sys/socket.h>
     81 #include <sys/socketvar.h>
     82 #include <sys/errno.h>
     83 #include <sys/time.h>
     84 #include <sys/kernel.h>
     85 #include <sys/syslog.h>
     86 #include <sys/proc.h>
     87 #include <sys/sysctl.h>
     88 #include <sys/cprng.h>
     89 #include <sys/percpu.h>
     90 
     91 #include <net/if.h>
     92 #include <net/if_types.h>
     93 #include <net/if_dl.h>
     94 #include <net/route.h>
     95 #include <net/pktqueue.h>
     96 #include <net/pfil.h>
     97 
     98 #include <netinet/in.h>
     99 #include <netinet/in_systm.h>
    100 #ifdef INET
    101 #include <netinet/ip.h>
    102 #include <netinet/ip_var.h>
    103 #include <netinet/ip_icmp.h>
    104 #endif /* INET */
    105 #include <netinet/ip6.h>
    106 #include <netinet/portalgo.h>
    107 #include <netinet6/in6_var.h>
    108 #include <netinet6/ip6_var.h>
    109 #include <netinet6/ip6_private.h>
    110 #include <netinet6/in6_pcb.h>
    111 #include <netinet/icmp6.h>
    112 #include <netinet6/scope6_var.h>
    113 #include <netinet6/in6_ifattach.h>
    114 #include <netinet6/nd6.h>
    115 
    116 #ifdef IPSEC
    117 #include <netipsec/ipsec.h>
    118 #include <netipsec/ipsec6.h>
    119 #include <netipsec/key.h>
    120 #endif /* IPSEC */
    121 
    122 #include <netinet6/ip6protosw.h>
    123 
    124 #include "faith.h"
    125 
    126 extern struct domain inet6domain;
    127 
    128 u_char ip6_protox[IPPROTO_MAX];
    129 pktqueue_t *ip6_pktq __read_mostly;
    130 
    131 pfil_head_t *inet6_pfil_hook;
    132 
    133 percpu_t *ip6stat_percpu;
    134 
    135 percpu_t *ip6_forward_rt_percpu __cacheline_aligned;
    136 
    137 static void ip6intr(void *);
    138 static void ip6_input(struct mbuf *, struct ifnet *);
    139 static bool ip6_badaddr(struct ip6_hdr *);
    140 static struct m_tag *ip6_setdstifaddr(struct mbuf *, const struct in6_ifaddr *);
    141 
    142 static struct m_tag *ip6_addaux(struct mbuf *);
    143 static struct m_tag *ip6_findaux(struct mbuf *);
    144 static void ip6_delaux(struct mbuf *);
    145 
    146 static int ip6_process_hopopts(struct mbuf *, u_int8_t *, int, u_int32_t *,
    147     u_int32_t *);
    148 static struct mbuf *ip6_pullexthdr(struct mbuf *, size_t, int);
    149 static void sysctl_net_inet6_ip6_setup(struct sysctllog **);
    150 
    151 #ifdef NET_MPSAFE
    152 #define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
    153 #define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
    154 #else
    155 #define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
    156 #define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
    157 #endif
    158 
    159 /*
    160  * IP6 initialization: fill in IP6 protocol switch table.
    161  * All protocols not implemented in kernel go to raw IP6 protocol handler.
    162  */
    163 void
    164 ip6_init(void)
    165 {
    166 	const struct ip6protosw *pr;
    167 	int i;
    168 
    169 	in6_init();
    170 
    171 	sysctl_net_inet6_ip6_setup(NULL);
    172 	pr = (const struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
    173 	if (pr == 0)
    174 		panic("ip6_init");
    175 	for (i = 0; i < IPPROTO_MAX; i++)
    176 		ip6_protox[i] = pr - inet6sw;
    177 	for (pr = (const struct ip6protosw *)inet6domain.dom_protosw;
    178 	    pr < (const struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
    179 		if (pr->pr_domain->dom_family == PF_INET6 &&
    180 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
    181 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
    182 
    183 	ip6_pktq = pktq_create(IFQ_MAXLEN, ip6intr, NULL);
    184 	KASSERT(ip6_pktq != NULL);
    185 
    186 	scope6_init();
    187 	addrsel_policy_init();
    188 	nd6_init();
    189 	frag6_init();
    190 
    191 #ifdef GATEWAY
    192 	ip6flow_init(ip6_hashsize);
    193 #endif
    194 	/* Register our Packet Filter hook. */
    195 	inet6_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET6);
    196 	KASSERT(inet6_pfil_hook != NULL);
    197 
    198 	ip6stat_percpu = percpu_alloc(sizeof(uint64_t) * IP6_NSTATS);
    199 	ip6_forward_rt_percpu = rtcache_percpu_alloc();
    200 }
    201 
    202 /*
    203  * IP6 input interrupt handling. Just pass the packet to ip6_input.
    204  */
    205 static void
    206 ip6intr(void *arg __unused)
    207 {
    208 	struct mbuf *m;
    209 
    210 	SOFTNET_KERNEL_LOCK_UNLESS_NET_MPSAFE();
    211 	while ((m = pktq_dequeue(ip6_pktq)) != NULL) {
    212 		struct psref psref;
    213 		struct ifnet *rcvif = m_get_rcvif_psref(m, &psref);
    214 
    215 		if (rcvif == NULL) {
    216 			m_freem(m);
    217 			continue;
    218 		}
    219 		/*
    220 		 * Drop the packet if IPv6 is disabled on the interface.
    221 		 */
    222 		if ((ND_IFINFO(rcvif)->flags & ND6_IFF_IFDISABLED)) {
    223 			m_put_rcvif_psref(rcvif, &psref);
    224 			m_freem(m);
    225 			continue;
    226 		}
    227 		ip6_input(m, rcvif);
    228 		m_put_rcvif_psref(rcvif, &psref);
    229 	}
    230 	SOFTNET_KERNEL_UNLOCK_UNLESS_NET_MPSAFE();
    231 }
    232 
    233 static void
    234 ip6_input(struct mbuf *m, struct ifnet *rcvif)
    235 {
    236 	struct ip6_hdr *ip6;
    237 	int hit, off = sizeof(struct ip6_hdr), nest;
    238 	u_int32_t plen;
    239 	u_int32_t rtalert = ~0;
    240 	int nxt, ours = 0, rh_present = 0, frg_present;
    241 	struct ifnet *deliverifp = NULL;
    242 	int srcrt = 0;
    243 	struct rtentry *rt = NULL;
    244 	union {
    245 		struct sockaddr		dst;
    246 		struct sockaddr_in6	dst6;
    247 	} u;
    248 	struct route *ro;
    249 
    250 	KASSERT(rcvif != NULL);
    251 
    252 	/*
    253 	 * make sure we don't have onion peering information into m_tag.
    254 	 */
    255 	ip6_delaux(m);
    256 
    257 	/*
    258 	 * mbuf statistics
    259 	 */
    260 	if (m->m_flags & M_EXT) {
    261 		if (m->m_next)
    262 			IP6_STATINC(IP6_STAT_MEXT2M);
    263 		else
    264 			IP6_STATINC(IP6_STAT_MEXT1);
    265 	} else {
    266 #define M2MMAX	32
    267 		if (m->m_next) {
    268 			if (m->m_flags & M_LOOP)
    269 			/*XXX*/	IP6_STATINC(IP6_STAT_M2M + lo0ifp->if_index);
    270 			else if (rcvif->if_index < M2MMAX)
    271 				IP6_STATINC(IP6_STAT_M2M + rcvif->if_index);
    272 			else
    273 				IP6_STATINC(IP6_STAT_M2M);
    274 		} else
    275 			IP6_STATINC(IP6_STAT_M1);
    276 #undef M2MMAX
    277 	}
    278 
    279 	in6_ifstat_inc(rcvif, ifs6_in_receive);
    280 	IP6_STATINC(IP6_STAT_TOTAL);
    281 
    282 	/*
    283 	 * If the IPv6 header is not aligned, slurp it up into a new
    284 	 * mbuf with space for link headers, in the event we forward
    285 	 * it.  Otherwise, if it is aligned, make sure the entire base
    286 	 * IPv6 header is in the first mbuf of the chain.
    287 	 */
    288 	if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
    289 		if ((m = m_copyup(m, sizeof(struct ip6_hdr),
    290 		    (max_linkhdr + 3) & ~3)) == NULL) {
    291 			/* XXXJRT new stat, please */
    292 			IP6_STATINC(IP6_STAT_TOOSMALL);
    293 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    294 			return;
    295 		}
    296 	} else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) {
    297 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    298 			IP6_STATINC(IP6_STAT_TOOSMALL);
    299 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    300 			return;
    301 		}
    302 	}
    303 
    304 	ip6 = mtod(m, struct ip6_hdr *);
    305 
    306 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
    307 		IP6_STATINC(IP6_STAT_BADVERS);
    308 		in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    309 		goto bad;
    310 	}
    311 
    312 	if (ip6_badaddr(ip6)) {
    313 		IP6_STATINC(IP6_STAT_BADSCOPE);
    314 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    315 		goto bad;
    316 	}
    317 
    318 	/*
    319 	 * Assume that we can create a fast-forward IP flow entry
    320 	 * based on this packet.
    321 	 */
    322 	m->m_flags |= M_CANFASTFWD;
    323 
    324 	/*
    325 	 * Run through list of hooks for input packets.  If there are any
    326 	 * filters which require that additional packets in the flow are
    327 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
    328 	 * Note that filters must _never_ set this flag, as another filter
    329 	 * in the list may have previously cleared it.
    330 	 *
    331 	 * Don't call hooks if the packet has already been processed by
    332 	 * IPsec (encapsulated, tunnel mode).
    333 	 */
    334 #if defined(IPSEC)
    335 	if (!ipsec_used || !ipsec_skip_pfil(m))
    336 #else
    337 	if (1)
    338 #endif
    339 	{
    340 		struct in6_addr odst;
    341 		int error;
    342 
    343 		odst = ip6->ip6_dst;
    344 		error = pfil_run_hooks(inet6_pfil_hook, &m, rcvif, PFIL_IN);
    345 		if (error != 0 || m == NULL) {
    346 			IP6_STATINC(IP6_STAT_PFILDROP_IN);
    347 			return;
    348 		}
    349 		if (m->m_len < sizeof(struct ip6_hdr)) {
    350 			if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
    351 				IP6_STATINC(IP6_STAT_TOOSMALL);
    352 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    353 				return;
    354 			}
    355 		}
    356 		ip6 = mtod(m, struct ip6_hdr *);
    357 		srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
    358 	}
    359 
    360 	IP6_STATINC(IP6_STAT_NXTHIST + ip6->ip6_nxt);
    361 
    362 #ifdef ALTQ
    363 	if (altq_input != NULL) {
    364 		SOFTNET_LOCK();
    365 		if ((*altq_input)(m, AF_INET6) == 0) {
    366 			SOFTNET_UNLOCK();
    367 			/* packet is dropped by traffic conditioner */
    368 			return;
    369 		}
    370 		SOFTNET_UNLOCK();
    371 	}
    372 #endif
    373 
    374 	/*
    375 	 * Disambiguate address scope zones (if there is ambiguity).
    376 	 * We first make sure that the original source or destination address
    377 	 * is not in our internal form for scoped addresses.  Such addresses
    378 	 * are not necessarily invalid spec-wise, but we cannot accept them due
    379 	 * to the usage conflict.
    380 	 * in6_setscope() then also checks and rejects the cases where src or
    381 	 * dst are the loopback address and the receiving interface
    382 	 * is not loopback.
    383 	 */
    384 	if (__predict_false(
    385 	    m_makewritable(&m, 0, sizeof(struct ip6_hdr), M_DONTWAIT)))
    386 		goto bad;
    387 	ip6 = mtod(m, struct ip6_hdr *);
    388 	if (in6_clearscope(&ip6->ip6_src) || in6_clearscope(&ip6->ip6_dst)) {
    389 		IP6_STATINC(IP6_STAT_BADSCOPE);	/* XXX */
    390 		goto bad;
    391 	}
    392 	if (in6_setscope(&ip6->ip6_src, rcvif, NULL) ||
    393 	    in6_setscope(&ip6->ip6_dst, rcvif, NULL)) {
    394 		IP6_STATINC(IP6_STAT_BADSCOPE);
    395 		goto bad;
    396 	}
    397 
    398 	ro = rtcache_percpu_getref(ip6_forward_rt_percpu);
    399 
    400 	/*
    401 	 * Multicast check
    402 	 */
    403 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    404 		bool ingroup;
    405 
    406 		in6_ifstat_inc(rcvif, ifs6_in_mcast);
    407 		/*
    408 		 * See if we belong to the destination multicast group on the
    409 		 * arrival interface.
    410 		 */
    411 		ingroup = in6_multi_group(&ip6->ip6_dst, rcvif);
    412 		if (ingroup) {
    413 			ours = 1;
    414 		} else if (!ip6_mrouter) {
    415 			uint64_t *ip6s = IP6_STAT_GETREF();
    416 			ip6s[IP6_STAT_NOTMEMBER]++;
    417 			ip6s[IP6_STAT_CANTFORWARD]++;
    418 			IP6_STAT_PUTREF();
    419 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    420 			goto bad_unref;
    421 		}
    422 		deliverifp = rcvif;
    423 		goto hbhcheck;
    424 	}
    425 
    426 	sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
    427 
    428 	/*
    429 	 * Unicast check
    430 	 */
    431 	rt = rtcache_lookup2(ro, &u.dst, 1, &hit);
    432 	if (hit)
    433 		IP6_STATINC(IP6_STAT_FORWARD_CACHEHIT);
    434 	else
    435 		IP6_STATINC(IP6_STAT_FORWARD_CACHEMISS);
    436 
    437 	/*
    438 	 * Accept the packet if the forwarding interface to the destination
    439 	 * (according to the routing table) is the loopback interface,
    440 	 * unless the associated route has a gateway.
    441 	 *
    442 	 * We don't explicitly match ip6_dst against an interface here. It
    443 	 * is already done in rtcache_lookup2: rt->rt_ifp->if_type will be
    444 	 * IFT_LOOP if the packet is for us.
    445 	 *
    446 	 * Note that this approach causes to accept a packet if there is a
    447 	 * route to the loopback interface for the destination of the packet.
    448 	 * But we think it's even useful in some situations, e.g. when using
    449 	 * a special daemon which wants to intercept the packet.
    450 	 */
    451 	if (rt != NULL &&
    452 	    (rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
    453 	    rt->rt_ifp->if_type == IFT_LOOP) {
    454 		struct in6_ifaddr *ia6 = (struct in6_ifaddr *)rt->rt_ifa;
    455 		int addrok;
    456 
    457 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
    458 			m->m_flags |= M_ANYCAST6;
    459 		/*
    460 		 * packets to a tentative, duplicated, or somehow invalid
    461 		 * address must not be accepted.
    462 		 */
    463 		if (ia6->ia6_flags & IN6_IFF_NOTREADY)
    464 			addrok = 0;
    465 		else if (ia6->ia6_flags & IN6_IFF_DETACHED &&
    466 		    !IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src))
    467 		{
    468 			/* Allow internal traffic to DETACHED addresses */
    469 			struct sockaddr_in6 sin6;
    470 			int s;
    471 
    472 			memset(&sin6, 0, sizeof(sin6));
    473 			sin6.sin6_family = AF_INET6;
    474 			sin6.sin6_len = sizeof(sin6);
    475 			sin6.sin6_addr = ip6->ip6_src;
    476 			s = pserialize_read_enter();
    477 			addrok = (ifa_ifwithaddr(sin6tosa(&sin6)) != NULL);
    478 			pserialize_read_exit(s);
    479 		} else
    480 			addrok = 1;
    481 		if (addrok) {
    482 			/* this address is ready */
    483 			ours = 1;
    484 			deliverifp = ia6->ia_ifp;	/* correct? */
    485 			goto hbhcheck;
    486 		} else {
    487 			/* address is not ready, so discard the packet. */
    488 			char ip6bufs[INET6_ADDRSTRLEN];
    489 			char ip6bufd[INET6_ADDRSTRLEN];
    490 			nd6log(LOG_INFO, "packet to an unready address %s->%s\n",
    491 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
    492 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
    493 
    494 			goto bad_unref;
    495 		}
    496 	}
    497 
    498 	/*
    499 	 * FAITH (Firewall Aided Internet Translator)
    500 	 */
    501 #if defined(NFAITH) && 0 < NFAITH
    502 	if (ip6_keepfaith) {
    503 		if (rt != NULL && rt->rt_ifp != NULL &&
    504 		    rt->rt_ifp->if_type == IFT_FAITH) {
    505 			/* XXX do we need more sanity checks? */
    506 			ours = 1;
    507 			deliverifp = rt->rt_ifp; /* faith */
    508 			goto hbhcheck;
    509 		}
    510 	}
    511 #endif
    512 
    513 	/*
    514 	 * Now there is no reason to process the packet if it's not our own
    515 	 * and we're not a router.
    516 	 */
    517 	if (!ip6_forwarding) {
    518 		IP6_STATINC(IP6_STAT_CANTFORWARD);
    519 		in6_ifstat_inc(rcvif, ifs6_in_discard);
    520 		goto bad_unref;
    521 	}
    522 
    523 hbhcheck:
    524 	/*
    525 	 * Record address information into m_tag, if we don't have one yet.
    526 	 * Note that we are unable to record it, if the address is not listed
    527 	 * as our interface address (e.g. multicast addresses, addresses
    528 	 * within FAITH prefixes and such).
    529 	 */
    530 	if (deliverifp && ip6_getdstifaddr(m) == NULL) {
    531 		struct in6_ifaddr *ia6;
    532 		int s = pserialize_read_enter();
    533 
    534 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    535 		/* Depends on ip6_setdstifaddr never sleep */
    536 		if (ia6 != NULL && ip6_setdstifaddr(m, ia6) == NULL) {
    537 			/*
    538 			 * XXX maybe we should drop the packet here,
    539 			 * as we could not provide enough information
    540 			 * to the upper layers.
    541 			 */
    542 		}
    543 		pserialize_read_exit(s);
    544 	}
    545 
    546 	/*
    547 	 * Process Hop-by-Hop options header if it's contained.
    548 	 * m may be modified in ip6_hopopts_input().
    549 	 * If a JumboPayload option is included, plen will also be modified.
    550 	 */
    551 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
    552 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
    553 		struct ip6_hbh *hbh;
    554 
    555 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
    556 			/* m already freed */
    557 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    558 			rtcache_unref(rt, ro);
    559 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    560 			return;
    561 		}
    562 
    563 		/* adjust pointer */
    564 		ip6 = mtod(m, struct ip6_hdr *);
    565 
    566 		/*
    567 		 * if the payload length field is 0 and the next header field
    568 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
    569 		 * option MUST be included.
    570 		 */
    571 		if (ip6->ip6_plen == 0 && plen == 0) {
    572 			/*
    573 			 * Note that if a valid jumbo payload option is
    574 			 * contained, ip6_hopopts_input() must set a valid
    575 			 * (non-zero) payload length to the variable plen.
    576 			 */
    577 			IP6_STATINC(IP6_STAT_BADOPTIONS);
    578 			in6_ifstat_inc(rcvif, ifs6_in_discard);
    579 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    580 			icmp6_error(m, ICMP6_PARAM_PROB,
    581 				    ICMP6_PARAMPROB_HEADER,
    582 				    (char *)&ip6->ip6_plen - (char *)ip6);
    583 			rtcache_unref(rt, ro);
    584 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    585 			return;
    586 		}
    587 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    588 			sizeof(struct ip6_hbh));
    589 		if (hbh == NULL) {
    590 			IP6_STATINC(IP6_STAT_TOOSHORT);
    591 			rtcache_unref(rt, ro);
    592 			rtcache_percpu_putref(ip6_forward_rt_percpu);
    593 			return;
    594 		}
    595 		KASSERT(IP6_HDR_ALIGNED_P(hbh));
    596 		nxt = hbh->ip6h_nxt;
    597 
    598 		/*
    599 		 * accept the packet if a router alert option is included
    600 		 * and we act as an IPv6 router.
    601 		 */
    602 		if (rtalert != ~0 && ip6_forwarding)
    603 			ours = 1;
    604 	} else
    605 		nxt = ip6->ip6_nxt;
    606 
    607 	/*
    608 	 * Check that the amount of data in the buffers is at least much as
    609 	 * the IPv6 header would have us expect. Trim mbufs if longer than we
    610 	 * expect. Drop packet if shorter than we expect.
    611 	 */
    612 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
    613 		IP6_STATINC(IP6_STAT_TOOSHORT);
    614 		in6_ifstat_inc(rcvif, ifs6_in_truncated);
    615 		goto bad_unref;
    616 	}
    617 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
    618 		if (m->m_len == m->m_pkthdr.len) {
    619 			m->m_len = sizeof(struct ip6_hdr) + plen;
    620 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
    621 		} else
    622 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
    623 	}
    624 
    625 	/*
    626 	 * Forward if desirable.
    627 	 */
    628 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
    629 		/*
    630 		 * If we are acting as a multicast router, all
    631 		 * incoming multicast packets are passed to the
    632 		 * kernel-level multicast forwarding function.
    633 		 * The packet is returned (relatively) intact; if
    634 		 * ip6_mforward() returns a non-zero value, the packet
    635 		 * must be discarded, else it may be accepted below.
    636 		 */
    637 		if (ip6_mrouter != NULL) {
    638 			int error;
    639 
    640 			SOFTNET_LOCK();
    641 			error = ip6_mforward(ip6, rcvif, m);
    642 			SOFTNET_UNLOCK();
    643 
    644 			if (error != 0) {
    645 				rtcache_unref(rt, ro);
    646 				rtcache_percpu_putref(ip6_forward_rt_percpu);
    647 				IP6_STATINC(IP6_STAT_CANTFORWARD);
    648 				goto bad;
    649 			}
    650 		}
    651 		if (!ours)
    652 			goto bad_unref;
    653 	} else if (!ours) {
    654 		rtcache_unref(rt, ro);
    655 		rtcache_percpu_putref(ip6_forward_rt_percpu);
    656 		ip6_forward(m, srcrt);
    657 		return;
    658 	}
    659 
    660 	ip6 = mtod(m, struct ip6_hdr *);
    661 
    662 	/*
    663 	 * Malicious party may be able to use IPv4 mapped addr to confuse
    664 	 * tcp/udp stack and bypass security checks (act as if it was from
    665 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
    666 	 *
    667 	 * For SIIT end node behavior, you may want to disable the check.
    668 	 * However, you will  become vulnerable to attacks using IPv4 mapped
    669 	 * source.
    670 	 */
    671 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    672 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    673 		IP6_STATINC(IP6_STAT_BADSCOPE);
    674 		in6_ifstat_inc(rcvif, ifs6_in_addrerr);
    675 		goto bad_unref;
    676 	}
    677 
    678 #ifdef IFA_STATS
    679 	if (deliverifp != NULL) {
    680 		struct in6_ifaddr *ia6;
    681 		int s = pserialize_read_enter();
    682 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
    683 		if (ia6)
    684 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
    685 		pserialize_read_exit(s);
    686 	}
    687 #endif
    688 	IP6_STATINC(IP6_STAT_DELIVERED);
    689 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
    690 	nest = 0;
    691 
    692 	if (rt != NULL) {
    693 		rtcache_unref(rt, ro);
    694 		rt = NULL;
    695 	}
    696 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    697 
    698 	rh_present = 0;
    699 	frg_present = 0;
    700 	while (nxt != IPPROTO_DONE) {
    701 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
    702 			IP6_STATINC(IP6_STAT_TOOMANYHDR);
    703 			in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    704 			goto bad;
    705 		}
    706 
    707 		M_VERIFY_PACKET(m);
    708 
    709 		/*
    710 		 * protection against faulty packet - there should be
    711 		 * more sanity checks in header chain processing.
    712 		 */
    713 		if (m->m_pkthdr.len < off) {
    714 			IP6_STATINC(IP6_STAT_TOOSHORT);
    715 			in6_ifstat_inc(rcvif, ifs6_in_truncated);
    716 			goto bad;
    717 		}
    718 
    719 		if (nxt == IPPROTO_ROUTING) {
    720 			if (rh_present++) {
    721 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    722 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    723 				goto bad;
    724 			}
    725 		} else if (nxt == IPPROTO_FRAGMENT) {
    726 			if (frg_present++) {
    727 				in6_ifstat_inc(rcvif, ifs6_in_hdrerr);
    728 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    729 				goto bad;
    730 			}
    731 		}
    732 
    733 #ifdef IPSEC
    734 		if (ipsec_used) {
    735 			/*
    736 			 * Enforce IPsec policy checking if we are seeing last
    737 			 * header. Note that we do not visit this with
    738 			 * protocols with pcb layer code - like udp/tcp/raw ip.
    739 			 */
    740 			if ((inet6sw[ip6_protox[nxt]].pr_flags
    741 			    & PR_LASTHDR) != 0) {
    742 				int error;
    743 
    744 				error = ipsec_ip_input(m, false);
    745 				if (error)
    746 					goto bad;
    747 			}
    748 		}
    749 #endif
    750 
    751 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
    752 	}
    753 	return;
    754 
    755 bad_unref:
    756 	rtcache_unref(rt, ro);
    757 	rtcache_percpu_putref(ip6_forward_rt_percpu);
    758 bad:
    759 	m_freem(m);
    760 	return;
    761 }
    762 
    763 static bool
    764 ip6_badaddr(struct ip6_hdr *ip6)
    765 {
    766 	/* Check against address spoofing/corruption. */
    767 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
    768 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
    769 		return true;
    770 	}
    771 
    772 	/*
    773 	 * The following check is not documented in specs.  A malicious
    774 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
    775 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
    776 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
    777 	 *
    778 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
    779 	 * support IPv4-less kernel compilation, we cannot support SIIT
    780 	 * environment at all.  So, it makes more sense for us to reject any
    781 	 * malicious packets for non-SIIT environment, than try to do a
    782 	 * partial support for SIIT environment.
    783 	 */
    784 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
    785 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
    786 		return true;
    787 	}
    788 
    789 	/*
    790 	 * Reject packets with IPv4-compatible IPv6 addresses (RFC4291).
    791 	 */
    792 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
    793 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
    794 		return true;
    795 	}
    796 
    797 	return false;
    798 }
    799 
    800 /*
    801  * set/grab in6_ifaddr correspond to IPv6 destination address.
    802  */
    803 static struct m_tag *
    804 ip6_setdstifaddr(struct mbuf *m, const struct in6_ifaddr *ia)
    805 {
    806 	struct m_tag *mtag;
    807 	struct ip6aux *ip6a;
    808 
    809 	mtag = ip6_addaux(m);
    810 	if (mtag == NULL)
    811 		return NULL;
    812 
    813 	ip6a = (struct ip6aux *)(mtag + 1);
    814 	if (in6_setscope(&ip6a->ip6a_src, ia->ia_ifp, &ip6a->ip6a_scope_id)) {
    815 		IP6_STATINC(IP6_STAT_BADSCOPE);
    816 		return NULL;
    817 	}
    818 
    819 	ip6a->ip6a_src = ia->ia_addr.sin6_addr;
    820 	ip6a->ip6a_flags = ia->ia6_flags;
    821 	return mtag;
    822 }
    823 
    824 const struct ip6aux *
    825 ip6_getdstifaddr(struct mbuf *m)
    826 {
    827 	struct m_tag *mtag;
    828 
    829 	mtag = ip6_findaux(m);
    830 	if (mtag != NULL)
    831 		return (struct ip6aux *)(mtag + 1);
    832 	else
    833 		return NULL;
    834 }
    835 
    836 /*
    837  * Hop-by-Hop options header processing. If a valid jumbo payload option is
    838  * included, the real payload length will be stored in plenp.
    839  *
    840  * rtalertp - XXX: should be stored more smart way
    841  */
    842 int
    843 ip6_hopopts_input(u_int32_t *plenp, u_int32_t *rtalertp,
    844 	struct mbuf **mp, int *offp)
    845 {
    846 	struct mbuf *m = *mp;
    847 	int off = *offp, hbhlen;
    848 	struct ip6_hbh *hbh;
    849 
    850 	/* validation of the length of the header */
    851 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
    852 	    sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
    853 	if (hbh == NULL) {
    854 		IP6_STATINC(IP6_STAT_TOOSHORT);
    855 		return -1;
    856 	}
    857 	hbhlen = (hbh->ip6h_len + 1) << 3;
    858 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
    859 	    hbhlen);
    860 	if (hbh == NULL) {
    861 		IP6_STATINC(IP6_STAT_TOOSHORT);
    862 		return -1;
    863 	}
    864 	KASSERT(IP6_HDR_ALIGNED_P(hbh));
    865 	off += hbhlen;
    866 	hbhlen -= sizeof(struct ip6_hbh);
    867 
    868 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
    869 	    hbhlen, rtalertp, plenp) < 0)
    870 		return -1;
    871 
    872 	*offp = off;
    873 	*mp = m;
    874 	return 0;
    875 }
    876 
    877 /*
    878  * Search header for all Hop-by-hop options and process each option.
    879  * This function is separate from ip6_hopopts_input() in order to
    880  * handle a case where the sending node itself process its hop-by-hop
    881  * options header. In such a case, the function is called from ip6_output().
    882  *
    883  * The function assumes that hbh header is located right after the IPv6 header
    884  * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
    885  * opthead + hbhlen is located in continuous memory region.
    886  */
    887 static int
    888 ip6_process_hopopts(struct mbuf *m, u_int8_t *opthead, int hbhlen,
    889 	u_int32_t *rtalertp, u_int32_t *plenp)
    890 {
    891 	struct ip6_hdr *ip6;
    892 	int optlen = 0;
    893 	u_int8_t *opt = opthead;
    894 	u_int16_t rtalert_val;
    895 	u_int32_t jumboplen;
    896 	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
    897 
    898 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
    899 		switch (*opt) {
    900 		case IP6OPT_PAD1:
    901 			optlen = 1;
    902 			break;
    903 		case IP6OPT_PADN:
    904 			if (hbhlen < IP6OPT_MINLEN) {
    905 				IP6_STATINC(IP6_STAT_TOOSMALL);
    906 				goto bad;
    907 			}
    908 			optlen = *(opt + 1) + 2;
    909 			break;
    910 		case IP6OPT_RTALERT:
    911 			/* XXX may need check for alignment */
    912 			if (hbhlen < IP6OPT_RTALERT_LEN) {
    913 				IP6_STATINC(IP6_STAT_TOOSMALL);
    914 				goto bad;
    915 			}
    916 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
    917 				/* XXX stat */
    918 				icmp6_error(m, ICMP6_PARAM_PROB,
    919 				    ICMP6_PARAMPROB_HEADER,
    920 				    erroff + opt + 1 - opthead);
    921 				return (-1);
    922 			}
    923 			optlen = IP6OPT_RTALERT_LEN;
    924 			memcpy((void *)&rtalert_val, (void *)(opt + 2), 2);
    925 			*rtalertp = ntohs(rtalert_val);
    926 			break;
    927 		case IP6OPT_JUMBO:
    928 			/* XXX may need check for alignment */
    929 			if (hbhlen < IP6OPT_JUMBO_LEN) {
    930 				IP6_STATINC(IP6_STAT_TOOSMALL);
    931 				goto bad;
    932 			}
    933 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
    934 				/* XXX stat */
    935 				icmp6_error(m, ICMP6_PARAM_PROB,
    936 				    ICMP6_PARAMPROB_HEADER,
    937 				    erroff + opt + 1 - opthead);
    938 				return (-1);
    939 			}
    940 			optlen = IP6OPT_JUMBO_LEN;
    941 
    942 			/*
    943 			 * IPv6 packets that have non 0 payload length
    944 			 * must not contain a jumbo payload option.
    945 			 */
    946 			ip6 = mtod(m, struct ip6_hdr *);
    947 			if (ip6->ip6_plen) {
    948 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    949 				icmp6_error(m, ICMP6_PARAM_PROB,
    950 				    ICMP6_PARAMPROB_HEADER,
    951 				    erroff + opt - opthead);
    952 				return (-1);
    953 			}
    954 
    955 			/*
    956 			 * We may see jumbolen in unaligned location, so
    957 			 * we'd need to perform memcpy().
    958 			 */
    959 			memcpy(&jumboplen, opt + 2, sizeof(jumboplen));
    960 			jumboplen = (u_int32_t)htonl(jumboplen);
    961 
    962 #if 1
    963 			/*
    964 			 * if there are multiple jumbo payload options,
    965 			 * *plenp will be non-zero and the packet will be
    966 			 * rejected.
    967 			 * the behavior may need some debate in ipngwg -
    968 			 * multiple options does not make sense, however,
    969 			 * there's no explicit mention in specification.
    970 			 */
    971 			if (*plenp != 0) {
    972 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    973 				icmp6_error(m, ICMP6_PARAM_PROB,
    974 				    ICMP6_PARAMPROB_HEADER,
    975 				    erroff + opt + 2 - opthead);
    976 				return (-1);
    977 			}
    978 #endif
    979 
    980 			/*
    981 			 * jumbo payload length must be larger than 65535.
    982 			 */
    983 			if (jumboplen <= IPV6_MAXPACKET) {
    984 				IP6_STATINC(IP6_STAT_BADOPTIONS);
    985 				icmp6_error(m, ICMP6_PARAM_PROB,
    986 				    ICMP6_PARAMPROB_HEADER,
    987 				    erroff + opt + 2 - opthead);
    988 				return (-1);
    989 			}
    990 			*plenp = jumboplen;
    991 
    992 			break;
    993 		default:		/* unknown option */
    994 			if (hbhlen < IP6OPT_MINLEN) {
    995 				IP6_STATINC(IP6_STAT_TOOSMALL);
    996 				goto bad;
    997 			}
    998 			optlen = ip6_unknown_opt(opt, m,
    999 			    erroff + opt - opthead);
   1000 			if (optlen == -1)
   1001 				return (-1);
   1002 			optlen += 2;
   1003 			break;
   1004 		}
   1005 	}
   1006 
   1007 	return (0);
   1008 
   1009   bad:
   1010 	m_freem(m);
   1011 	return (-1);
   1012 }
   1013 
   1014 /*
   1015  * Unknown option processing.
   1016  * The third argument `off' is the offset from the IPv6 header to the option,
   1017  * which is necessary if the IPv6 header the and option header and IPv6 header
   1018  * is not continuous in order to return an ICMPv6 error.
   1019  */
   1020 int
   1021 ip6_unknown_opt(u_int8_t *optp, struct mbuf *m, int off)
   1022 {
   1023 	struct ip6_hdr *ip6;
   1024 
   1025 	switch (IP6OPT_TYPE(*optp)) {
   1026 	case IP6OPT_TYPE_SKIP: /* ignore the option */
   1027 		return ((int)*(optp + 1));
   1028 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
   1029 		m_freem(m);
   1030 		return (-1);
   1031 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
   1032 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1033 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
   1034 		return (-1);
   1035 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
   1036 		IP6_STATINC(IP6_STAT_BADOPTIONS);
   1037 		ip6 = mtod(m, struct ip6_hdr *);
   1038 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
   1039 		    (m->m_flags & (M_BCAST|M_MCAST)))
   1040 			m_freem(m);
   1041 		else
   1042 			icmp6_error(m, ICMP6_PARAM_PROB,
   1043 				    ICMP6_PARAMPROB_OPTION, off);
   1044 		return (-1);
   1045 	}
   1046 
   1047 	m_freem(m);		/* XXX: NOTREACHED */
   1048 	return (-1);
   1049 }
   1050 
   1051 void
   1052 ip6_savecontrol(struct in6pcb *in6p, struct mbuf **mp,
   1053 	struct ip6_hdr *ip6, struct mbuf *m)
   1054 {
   1055 	struct socket *so = in6p->in6p_socket;
   1056 #ifdef RFC2292
   1057 #define IS2292(x, y)	((in6p->in6p_flags & IN6P_RFC2292) ? (x) : (y))
   1058 #else
   1059 #define IS2292(x, y)	(y)
   1060 #endif
   1061 
   1062 	KASSERT(m->m_flags & M_PKTHDR);
   1063 
   1064 	if (SOOPT_TIMESTAMP(so->so_options))
   1065 		mp = sbsavetimestamp(so->so_options, mp);
   1066 
   1067 	/* some OSes call this logic with IPv4 packet, for SO_TIMESTAMP */
   1068 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION)
   1069 		return;
   1070 
   1071 	/* RFC 2292 sec. 5 */
   1072 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
   1073 		struct in6_pktinfo pi6;
   1074 
   1075 		memcpy(&pi6.ipi6_addr, &ip6->ip6_dst, sizeof(struct in6_addr));
   1076 		in6_clearscope(&pi6.ipi6_addr);	/* XXX */
   1077 		pi6.ipi6_ifindex = m->m_pkthdr.rcvif_index;
   1078 		*mp = sbcreatecontrol(&pi6, sizeof(pi6),
   1079 		    IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6);
   1080 		if (*mp)
   1081 			mp = &(*mp)->m_next;
   1082 	}
   1083 
   1084 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
   1085 		int hlim = ip6->ip6_hlim & 0xff;
   1086 
   1087 		*mp = sbcreatecontrol(&hlim, sizeof(hlim),
   1088 		    IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6);
   1089 		if (*mp)
   1090 			mp = &(*mp)->m_next;
   1091 	}
   1092 
   1093 	if ((in6p->in6p_flags & IN6P_TCLASS) != 0) {
   1094 		u_int32_t flowinfo;
   1095 		int tclass;
   1096 
   1097 		flowinfo = (u_int32_t)ntohl(ip6->ip6_flow & IPV6_FLOWINFO_MASK);
   1098 		flowinfo >>= 20;
   1099 
   1100 		tclass = flowinfo & 0xff;
   1101 		*mp = sbcreatecontrol(&tclass, sizeof(tclass),
   1102 		    IPV6_TCLASS, IPPROTO_IPV6);
   1103 
   1104 		if (*mp)
   1105 			mp = &(*mp)->m_next;
   1106 	}
   1107 
   1108 	/*
   1109 	 * IPV6_HOPOPTS socket option.  Recall that we required super-user
   1110 	 * privilege for the option (see ip6_ctloutput), but it might be too
   1111 	 * strict, since there might be some hop-by-hop options which can be
   1112 	 * returned to normal user.
   1113 	 * See also RFC3542 section 8 (or RFC2292 section 6).
   1114 	 */
   1115 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0) {
   1116 		/*
   1117 		 * Check if a hop-by-hop options header is contatined in the
   1118 		 * received packet, and if so, store the options as ancillary
   1119 		 * data. Note that a hop-by-hop options header must be
   1120 		 * just after the IPv6 header, which fact is assured through
   1121 		 * the IPv6 input processing.
   1122 		 */
   1123 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1124 		if (xip6->ip6_nxt == IPPROTO_HOPOPTS) {
   1125 			struct ip6_hbh *hbh;
   1126 			int hbhlen;
   1127 			struct mbuf *ext;
   1128 
   1129 			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
   1130 			    xip6->ip6_nxt);
   1131 			if (ext == NULL) {
   1132 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1133 				return;
   1134 			}
   1135 			hbh = mtod(ext, struct ip6_hbh *);
   1136 			hbhlen = (hbh->ip6h_len + 1) << 3;
   1137 			if (hbhlen != ext->m_len) {
   1138 				m_freem(ext);
   1139 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1140 				return;
   1141 			}
   1142 
   1143 			/*
   1144 			 * XXX: We copy whole the header even if a jumbo
   1145 			 * payload option is included, which option is to
   1146 			 * be removed before returning in the RFC 2292.
   1147 			 * Note: this constraint is removed in RFC3542.
   1148 			 */
   1149 			*mp = sbcreatecontrol(hbh, hbhlen,
   1150 			    IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS),
   1151 			    IPPROTO_IPV6);
   1152 			if (*mp)
   1153 				mp = &(*mp)->m_next;
   1154 			m_freem(ext);
   1155 		}
   1156 	}
   1157 
   1158 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
   1159 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
   1160 		struct ip6_hdr *xip6 = mtod(m, struct ip6_hdr *);
   1161 		int nxt = xip6->ip6_nxt, off = sizeof(struct ip6_hdr);
   1162 
   1163 		/*
   1164 		 * Search for destination options headers or routing
   1165 		 * header(s) through the header chain, and stores each
   1166 		 * header as ancillary data.
   1167 		 * Note that the order of the headers remains in
   1168 		 * the chain of ancillary data.
   1169 		 */
   1170 		for (;;) {	/* is explicit loop prevention necessary? */
   1171 			struct ip6_ext *ip6e = NULL;
   1172 			int elen;
   1173 			struct mbuf *ext = NULL;
   1174 
   1175 			/*
   1176 			 * if it is not an extension header, don't try to
   1177 			 * pull it from the chain.
   1178 			 */
   1179 			switch (nxt) {
   1180 			case IPPROTO_DSTOPTS:
   1181 			case IPPROTO_ROUTING:
   1182 			case IPPROTO_HOPOPTS:
   1183 			case IPPROTO_AH: /* is it possible? */
   1184 				break;
   1185 			default:
   1186 				goto loopend;
   1187 			}
   1188 
   1189 			ext = ip6_pullexthdr(m, off, nxt);
   1190 			if (ext == NULL) {
   1191 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1192 				return;
   1193 			}
   1194 			ip6e = mtod(ext, struct ip6_ext *);
   1195 			if (nxt == IPPROTO_AH)
   1196 				elen = (ip6e->ip6e_len + 2) << 2;
   1197 			else
   1198 				elen = (ip6e->ip6e_len + 1) << 3;
   1199 			if (elen != ext->m_len) {
   1200 				m_freem(ext);
   1201 				IP6_STATINC(IP6_STAT_TOOSHORT);
   1202 				return;
   1203 			}
   1204 			KASSERT(IP6_HDR_ALIGNED_P(ip6e));
   1205 
   1206 			switch (nxt) {
   1207 			case IPPROTO_DSTOPTS:
   1208 				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
   1209 					break;
   1210 
   1211 				*mp = sbcreatecontrol(ip6e, elen,
   1212 				    IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS),
   1213 				    IPPROTO_IPV6);
   1214 				if (*mp)
   1215 					mp = &(*mp)->m_next;
   1216 				break;
   1217 
   1218 			case IPPROTO_ROUTING:
   1219 				if (!(in6p->in6p_flags & IN6P_RTHDR))
   1220 					break;
   1221 
   1222 				*mp = sbcreatecontrol(ip6e, elen,
   1223 				    IS2292(IPV6_2292RTHDR, IPV6_RTHDR),
   1224 				    IPPROTO_IPV6);
   1225 				if (*mp)
   1226 					mp = &(*mp)->m_next;
   1227 				break;
   1228 
   1229 			case IPPROTO_HOPOPTS:
   1230 			case IPPROTO_AH: /* is it possible? */
   1231 				break;
   1232 
   1233 			default:
   1234 				/*
   1235 			 	 * other cases have been filtered in the above.
   1236 				 * none will visit this case.  here we supply
   1237 				 * the code just in case (nxt overwritten or
   1238 				 * other cases).
   1239 				 */
   1240 				m_freem(ext);
   1241 				goto loopend;
   1242 
   1243 			}
   1244 
   1245 			/* proceed with the next header. */
   1246 			off += elen;
   1247 			nxt = ip6e->ip6e_nxt;
   1248 			ip6e = NULL;
   1249 			m_freem(ext);
   1250 			ext = NULL;
   1251 		}
   1252 	  loopend:
   1253 	  	;
   1254 	}
   1255 }
   1256 #undef IS2292
   1257 
   1258 
   1259 void
   1260 ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst,
   1261     uint32_t *mtu)
   1262 {
   1263 	struct socket *so;
   1264 	struct mbuf *m_mtu;
   1265 	struct ip6_mtuinfo mtuctl;
   1266 
   1267 	so = in6p->in6p_socket;
   1268 
   1269 	if (mtu == NULL)
   1270 		return;
   1271 
   1272 	KASSERT(so != NULL);
   1273 
   1274 	memset(&mtuctl, 0, sizeof(mtuctl));	/* zero-clear for safety */
   1275 	mtuctl.ip6m_mtu = *mtu;
   1276 	mtuctl.ip6m_addr = *dst;
   1277 	if (sa6_recoverscope(&mtuctl.ip6m_addr))
   1278 		return;
   1279 
   1280 	if ((m_mtu = sbcreatecontrol(&mtuctl, sizeof(mtuctl),
   1281 	    IPV6_PATHMTU, IPPROTO_IPV6)) == NULL)
   1282 		return;
   1283 
   1284 	if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu)
   1285 	    == 0) {
   1286 		soroverflow(so);
   1287 		m_freem(m_mtu);
   1288 	} else
   1289 		sorwakeup(so);
   1290 
   1291 	return;
   1292 }
   1293 
   1294 /*
   1295  * pull single extension header from mbuf chain.  returns single mbuf that
   1296  * contains the result, or NULL on error.
   1297  */
   1298 static struct mbuf *
   1299 ip6_pullexthdr(struct mbuf *m, size_t off, int nxt)
   1300 {
   1301 	struct ip6_ext ip6e;
   1302 	size_t elen;
   1303 	struct mbuf *n;
   1304 
   1305 	if (off + sizeof(ip6e) > m->m_pkthdr.len)
   1306 		return NULL;
   1307 
   1308 	m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1309 	if (nxt == IPPROTO_AH)
   1310 		elen = (ip6e.ip6e_len + 2) << 2;
   1311 	else
   1312 		elen = (ip6e.ip6e_len + 1) << 3;
   1313 
   1314 	if (off + elen > m->m_pkthdr.len)
   1315 		return NULL;
   1316 
   1317 	MGET(n, M_DONTWAIT, MT_DATA);
   1318 	if (n && elen >= MLEN) {
   1319 		MCLGET(n, M_DONTWAIT);
   1320 		if ((n->m_flags & M_EXT) == 0) {
   1321 			m_free(n);
   1322 			n = NULL;
   1323 		}
   1324 	}
   1325 	if (!n)
   1326 		return NULL;
   1327 
   1328 	n->m_len = 0;
   1329 	if (elen >= M_TRAILINGSPACE(n)) {
   1330 		m_free(n);
   1331 		return NULL;
   1332 	}
   1333 
   1334 	m_copydata(m, off, elen, mtod(n, void *));
   1335 	n->m_len = elen;
   1336 	return n;
   1337 }
   1338 
   1339 /*
   1340  * Get offset to the previous header followed by the header
   1341  * currently processed.
   1342  */
   1343 int
   1344 ip6_get_prevhdr(struct mbuf *m, int off)
   1345 {
   1346 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1347 
   1348 	if (off == sizeof(struct ip6_hdr)) {
   1349 		return offsetof(struct ip6_hdr, ip6_nxt);
   1350 	} else if (off < sizeof(struct ip6_hdr)) {
   1351 		panic("%s: off < sizeof(struct ip6_hdr)", __func__);
   1352 	} else {
   1353 		int len, nlen, nxt;
   1354 		struct ip6_ext ip6e;
   1355 
   1356 		nxt = ip6->ip6_nxt;
   1357 		len = sizeof(struct ip6_hdr);
   1358 		nlen = 0;
   1359 		while (len < off) {
   1360 			m_copydata(m, len, sizeof(ip6e), &ip6e);
   1361 
   1362 			switch (nxt) {
   1363 			case IPPROTO_FRAGMENT:
   1364 				nlen = sizeof(struct ip6_frag);
   1365 				break;
   1366 			case IPPROTO_AH:
   1367 				nlen = (ip6e.ip6e_len + 2) << 2;
   1368 				break;
   1369 			default:
   1370 				nlen = (ip6e.ip6e_len + 1) << 3;
   1371 				break;
   1372 			}
   1373 			len += nlen;
   1374 			nxt = ip6e.ip6e_nxt;
   1375 		}
   1376 
   1377 		return (len - nlen);
   1378 	}
   1379 }
   1380 
   1381 /*
   1382  * get next header offset.  m will be retained.
   1383  */
   1384 int
   1385 ip6_nexthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1386 {
   1387 	struct ip6_hdr ip6;
   1388 	struct ip6_ext ip6e;
   1389 	struct ip6_frag fh;
   1390 
   1391 	/* just in case */
   1392 	if (m == NULL)
   1393 		panic("%s: m == NULL", __func__);
   1394 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
   1395 		return -1;
   1396 
   1397 	switch (proto) {
   1398 	case IPPROTO_IPV6:
   1399 		/* do not chase beyond intermediate IPv6 headers */
   1400 		if (off != 0)
   1401 			return -1;
   1402 		if (m->m_pkthdr.len < off + sizeof(ip6))
   1403 			return -1;
   1404 		m_copydata(m, off, sizeof(ip6), (void *)&ip6);
   1405 		if (nxtp)
   1406 			*nxtp = ip6.ip6_nxt;
   1407 		off += sizeof(ip6);
   1408 		return off;
   1409 
   1410 	case IPPROTO_FRAGMENT:
   1411 		/*
   1412 		 * terminate parsing if it is not the first fragment,
   1413 		 * it does not make sense to parse through it.
   1414 		 */
   1415 		if (m->m_pkthdr.len < off + sizeof(fh))
   1416 			return -1;
   1417 		m_copydata(m, off, sizeof(fh), (void *)&fh);
   1418 		if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0)
   1419 			return -1;
   1420 		if (nxtp)
   1421 			*nxtp = fh.ip6f_nxt;
   1422 		off += sizeof(struct ip6_frag);
   1423 		return off;
   1424 
   1425 	case IPPROTO_AH:
   1426 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1427 			return -1;
   1428 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1429 		if (nxtp)
   1430 			*nxtp = ip6e.ip6e_nxt;
   1431 		off += (ip6e.ip6e_len + 2) << 2;
   1432 		if (m->m_pkthdr.len < off)
   1433 			return -1;
   1434 		return off;
   1435 
   1436 	case IPPROTO_HOPOPTS:
   1437 	case IPPROTO_ROUTING:
   1438 	case IPPROTO_DSTOPTS:
   1439 		if (m->m_pkthdr.len < off + sizeof(ip6e))
   1440 			return -1;
   1441 		m_copydata(m, off, sizeof(ip6e), (void *)&ip6e);
   1442 		if (nxtp)
   1443 			*nxtp = ip6e.ip6e_nxt;
   1444 		off += (ip6e.ip6e_len + 1) << 3;
   1445 		if (m->m_pkthdr.len < off)
   1446 			return -1;
   1447 		return off;
   1448 
   1449 	case IPPROTO_NONE:
   1450 	case IPPROTO_ESP:
   1451 	case IPPROTO_IPCOMP:
   1452 		/* give up */
   1453 		return -1;
   1454 
   1455 	default:
   1456 		return -1;
   1457 	}
   1458 }
   1459 
   1460 /*
   1461  * get offset for the last header in the chain.  m will be kept untainted.
   1462  */
   1463 int
   1464 ip6_lasthdr(struct mbuf *m, int off, int proto, int *nxtp)
   1465 {
   1466 	int newoff;
   1467 	int nxt;
   1468 
   1469 	if (!nxtp) {
   1470 		nxt = -1;
   1471 		nxtp = &nxt;
   1472 	}
   1473 	for (;;) {
   1474 		newoff = ip6_nexthdr(m, off, proto, nxtp);
   1475 		if (newoff < 0)
   1476 			return off;
   1477 		else if (newoff < off)
   1478 			return -1;	/* invalid */
   1479 		else if (newoff == off)
   1480 			return newoff;
   1481 
   1482 		off = newoff;
   1483 		proto = *nxtp;
   1484 	}
   1485 }
   1486 
   1487 static struct m_tag *
   1488 ip6_addaux(struct mbuf *m)
   1489 {
   1490 	struct m_tag *mtag;
   1491 
   1492 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1493 	if (!mtag) {
   1494 		mtag = m_tag_get(PACKET_TAG_INET6, sizeof(struct ip6aux),
   1495 		    M_NOWAIT);
   1496 		if (mtag) {
   1497 			m_tag_prepend(m, mtag);
   1498 			memset(mtag + 1, 0, sizeof(struct ip6aux));
   1499 		}
   1500 	}
   1501 	return mtag;
   1502 }
   1503 
   1504 static struct m_tag *
   1505 ip6_findaux(struct mbuf *m)
   1506 {
   1507 	struct m_tag *mtag;
   1508 
   1509 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1510 	return mtag;
   1511 }
   1512 
   1513 static void
   1514 ip6_delaux(struct mbuf *m)
   1515 {
   1516 	struct m_tag *mtag;
   1517 
   1518 	mtag = m_tag_find(m, PACKET_TAG_INET6);
   1519 	if (mtag)
   1520 		m_tag_delete(m, mtag);
   1521 }
   1522 
   1523 /*
   1524  * System control for IP6
   1525  */
   1526 
   1527 const u_char inet6ctlerrmap[PRC_NCMDS] = {
   1528 	0,		0,		0,		0,
   1529 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
   1530 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
   1531 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
   1532 	0,		0,		0,		0,
   1533 	ENOPROTOOPT
   1534 };
   1535 
   1536 extern int sysctl_net_inet6_addrctlpolicy(SYSCTLFN_ARGS);
   1537 
   1538 static int
   1539 sysctl_net_inet6_ip6_stats(SYSCTLFN_ARGS)
   1540 {
   1541 
   1542 	return (NETSTAT_SYSCTL(ip6stat_percpu, IP6_NSTATS));
   1543 }
   1544 
   1545 static void
   1546 sysctl_net_inet6_ip6_setup(struct sysctllog **clog)
   1547 {
   1548 
   1549 	sysctl_createv(clog, 0, NULL, NULL,
   1550 		       CTLFLAG_PERMANENT,
   1551 		       CTLTYPE_NODE, "inet6",
   1552 		       SYSCTL_DESCR("PF_INET6 related settings"),
   1553 		       NULL, 0, NULL, 0,
   1554 		       CTL_NET, PF_INET6, CTL_EOL);
   1555 	sysctl_createv(clog, 0, NULL, NULL,
   1556 		       CTLFLAG_PERMANENT,
   1557 		       CTLTYPE_NODE, "ip6",
   1558 		       SYSCTL_DESCR("IPv6 related settings"),
   1559 		       NULL, 0, NULL, 0,
   1560 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_EOL);
   1561 
   1562 	sysctl_createv(clog, 0, NULL, NULL,
   1563 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1564 		       CTLTYPE_INT, "forwarding",
   1565 		       SYSCTL_DESCR("Enable forwarding of INET6 datagrams"),
   1566 		       NULL, 0, &ip6_forwarding, 0,
   1567 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1568 		       IPV6CTL_FORWARDING, CTL_EOL);
   1569 	sysctl_createv(clog, 0, NULL, NULL,
   1570 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1571 		       CTLTYPE_INT, "redirect",
   1572 		       SYSCTL_DESCR("Enable sending of ICMPv6 redirect messages"),
   1573 		       NULL, 0, &ip6_sendredirects, 0,
   1574 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1575 		       IPV6CTL_SENDREDIRECTS, CTL_EOL);
   1576 	sysctl_createv(clog, 0, NULL, NULL,
   1577 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1578 		       CTLTYPE_INT, "hlim",
   1579 		       SYSCTL_DESCR("Hop limit for an INET6 datagram"),
   1580 		       NULL, 0, &ip6_defhlim, 0,
   1581 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1582 		       IPV6CTL_DEFHLIM, CTL_EOL);
   1583 	sysctl_createv(clog, 0, NULL, NULL,
   1584 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1585 		       CTLTYPE_INT, "maxfragpackets",
   1586 		       SYSCTL_DESCR("Maximum number of fragments to buffer "
   1587 				    "for reassembly"),
   1588 		       NULL, 0, &ip6_maxfragpackets, 0,
   1589 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1590 		       IPV6CTL_MAXFRAGPACKETS, CTL_EOL);
   1591 	sysctl_createv(clog, 0, NULL, NULL,
   1592 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1593 		       CTLTYPE_INT, "keepfaith",
   1594 		       SYSCTL_DESCR("Activate faith interface"),
   1595 		       NULL, 0, &ip6_keepfaith, 0,
   1596 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1597 		       IPV6CTL_KEEPFAITH, CTL_EOL);
   1598 	sysctl_createv(clog, 0, NULL, NULL,
   1599 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1600 		       CTLTYPE_INT, "log_interval",
   1601 		       SYSCTL_DESCR("Minimum interval between logging "
   1602 				    "unroutable packets"),
   1603 		       NULL, 0, &ip6_log_interval, 0,
   1604 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1605 		       IPV6CTL_LOG_INTERVAL, CTL_EOL);
   1606 	sysctl_createv(clog, 0, NULL, NULL,
   1607 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1608 		       CTLTYPE_INT, "hdrnestlimit",
   1609 		       SYSCTL_DESCR("Maximum number of nested IPv6 headers"),
   1610 		       NULL, 0, &ip6_hdrnestlimit, 0,
   1611 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1612 		       IPV6CTL_HDRNESTLIMIT, CTL_EOL);
   1613 	sysctl_createv(clog, 0, NULL, NULL,
   1614 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1615 		       CTLTYPE_INT, "dad_count",
   1616 		       SYSCTL_DESCR("Number of Duplicate Address Detection "
   1617 				    "probes to send"),
   1618 		       NULL, 0, &ip6_dad_count, 0,
   1619 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1620 		       IPV6CTL_DAD_COUNT, CTL_EOL);
   1621 	sysctl_createv(clog, 0, NULL, NULL,
   1622 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1623 		       CTLTYPE_INT, "auto_flowlabel",
   1624 		       SYSCTL_DESCR("Assign random IPv6 flow labels"),
   1625 		       NULL, 0, &ip6_auto_flowlabel, 0,
   1626 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1627 		       IPV6CTL_AUTO_FLOWLABEL, CTL_EOL);
   1628 	sysctl_createv(clog, 0, NULL, NULL,
   1629 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1630 		       CTLTYPE_INT, "defmcasthlim",
   1631 		       SYSCTL_DESCR("Default multicast hop limit"),
   1632 		       NULL, 0, &ip6_defmcasthlim, 0,
   1633 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1634 		       IPV6CTL_DEFMCASTHLIM, CTL_EOL);
   1635 	sysctl_createv(clog, 0, NULL, NULL,
   1636 		       CTLFLAG_PERMANENT,
   1637 		       CTLTYPE_STRING, "kame_version",
   1638 		       SYSCTL_DESCR("KAME Version"),
   1639 		       NULL, 0, __UNCONST(__KAME_VERSION), 0,
   1640 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1641 		       IPV6CTL_KAME_VERSION, CTL_EOL);
   1642 	sysctl_createv(clog, 0, NULL, NULL,
   1643 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1644 		       CTLTYPE_INT, "use_deprecated",
   1645 		       SYSCTL_DESCR("Allow use of deprecated addresses as "
   1646 				    "source addresses"),
   1647 		       NULL, 0, &ip6_use_deprecated, 0,
   1648 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1649 		       IPV6CTL_USE_DEPRECATED, CTL_EOL);
   1650 	sysctl_createv(clog, 0, NULL, NULL,
   1651 		       CTLFLAG_PERMANENT
   1652 #ifndef INET6_BINDV6ONLY
   1653 		       |CTLFLAG_READWRITE,
   1654 #endif
   1655 		       CTLTYPE_INT, "v6only",
   1656 		       SYSCTL_DESCR("Disallow PF_INET6 sockets from connecting "
   1657 				    "to PF_INET sockets"),
   1658 		       NULL, 0, &ip6_v6only, 0,
   1659 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1660 		       IPV6CTL_V6ONLY, CTL_EOL);
   1661 	sysctl_createv(clog, 0, NULL, NULL,
   1662 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1663 		       CTLTYPE_INT, "anonportmin",
   1664 		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
   1665 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmin, 0,
   1666 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1667 		       IPV6CTL_ANONPORTMIN, CTL_EOL);
   1668 	sysctl_createv(clog, 0, NULL, NULL,
   1669 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1670 		       CTLTYPE_INT, "anonportmax",
   1671 		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
   1672 		       sysctl_net_inet_ip_ports, 0, &ip6_anonportmax, 0,
   1673 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1674 		       IPV6CTL_ANONPORTMAX, CTL_EOL);
   1675 #ifndef IPNOPRIVPORTS
   1676 	sysctl_createv(clog, 0, NULL, NULL,
   1677 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1678 		       CTLTYPE_INT, "lowportmin",
   1679 		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
   1680 				    "to assign"),
   1681 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmin, 0,
   1682 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1683 		       IPV6CTL_LOWPORTMIN, CTL_EOL);
   1684 	sysctl_createv(clog, 0, NULL, NULL,
   1685 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1686 		       CTLTYPE_INT, "lowportmax",
   1687 		       SYSCTL_DESCR("Highest privileged ephemeral port number "
   1688 				    "to assign"),
   1689 		       sysctl_net_inet_ip_ports, 0, &ip6_lowportmax, 0,
   1690 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1691 		       IPV6CTL_LOWPORTMAX, CTL_EOL);
   1692 #endif /* IPNOPRIVPORTS */
   1693 	sysctl_createv(clog, 0, NULL, NULL,
   1694 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1695 		       CTLTYPE_INT, "auto_linklocal",
   1696 		       SYSCTL_DESCR("Default value of per-interface flag for "
   1697 		                    "adding an IPv6 link-local address to "
   1698 				    "interfaces when attached"),
   1699 		       NULL, 0, &ip6_auto_linklocal, 0,
   1700 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1701 		       IPV6CTL_AUTO_LINKLOCAL, CTL_EOL);
   1702 	sysctl_createv(clog, 0, NULL, NULL,
   1703 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
   1704 		       CTLTYPE_STRUCT, "addctlpolicy",
   1705 		       SYSCTL_DESCR("Return the current address control"
   1706 			   " policy"),
   1707 		       sysctl_net_inet6_addrctlpolicy, 0, NULL, 0,
   1708 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1709 		       IPV6CTL_ADDRCTLPOLICY, CTL_EOL);
   1710 	sysctl_createv(clog, 0, NULL, NULL,
   1711 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1712 		       CTLTYPE_INT, "prefer_tempaddr",
   1713 		       SYSCTL_DESCR("Prefer temporary address as source "
   1714 		                    "address"),
   1715 		       NULL, 0, &ip6_prefer_tempaddr, 0,
   1716 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1717 		       CTL_CREATE, CTL_EOL);
   1718 	sysctl_createv(clog, 0, NULL, NULL,
   1719 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1720 		       CTLTYPE_INT, "maxfrags",
   1721 		       SYSCTL_DESCR("Maximum fragments in reassembly queue"),
   1722 		       NULL, 0, &ip6_maxfrags, 0,
   1723 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1724 		       IPV6CTL_MAXFRAGS, CTL_EOL);
   1725 	sysctl_createv(clog, 0, NULL, NULL,
   1726 		       CTLFLAG_PERMANENT,
   1727 		       CTLTYPE_STRUCT, "stats",
   1728 		       SYSCTL_DESCR("IPv6 statistics"),
   1729 		       sysctl_net_inet6_ip6_stats, 0, NULL, 0,
   1730 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1731 		       IPV6CTL_STATS, CTL_EOL);
   1732 	sysctl_createv(clog, 0, NULL, NULL,
   1733 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1734 		       CTLTYPE_INT, "use_defaultzone",
   1735 		       SYSCTL_DESCR("Whether to use the default scope zones"),
   1736 		       NULL, 0, &ip6_use_defzone, 0,
   1737 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1738 		       IPV6CTL_USE_DEFAULTZONE, CTL_EOL);
   1739 	sysctl_createv(clog, 0, NULL, NULL,
   1740 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1741 		       CTLTYPE_INT, "mcast_pmtu",
   1742 		       SYSCTL_DESCR("Enable pMTU discovery for multicast packet"),
   1743 		       NULL, 0, &ip6_mcast_pmtu, 0,
   1744 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1745 		       CTL_CREATE, CTL_EOL);
   1746 	/* anonportalgo RFC6056 subtree */
   1747 	const struct sysctlnode *portalgo_node;
   1748 	sysctl_createv(clog, 0, NULL, &portalgo_node,
   1749 		       CTLFLAG_PERMANENT,
   1750 		       CTLTYPE_NODE, "anonportalgo",
   1751 		       SYSCTL_DESCR("Anonymous port algorithm selection (RFC 6056)"),
   1752 	    	       NULL, 0, NULL, 0,
   1753 		       CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL);
   1754 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1755 		       CTLFLAG_PERMANENT,
   1756 		       CTLTYPE_STRING, "available",
   1757 		       SYSCTL_DESCR("available algorithms"),
   1758 		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
   1759 		       CTL_CREATE, CTL_EOL);
   1760 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1761 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1762 		       CTLTYPE_STRING, "selected",
   1763 		       SYSCTL_DESCR("selected algorithm"),
   1764 	               sysctl_portalgo_selected6, 0, NULL, PORTALGO_MAXLEN,
   1765 		       CTL_CREATE, CTL_EOL);
   1766 	sysctl_createv(clog, 0, &portalgo_node, NULL,
   1767 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1768 		       CTLTYPE_STRUCT, "reserve",
   1769 		       SYSCTL_DESCR("bitmap of reserved ports"),
   1770 		       sysctl_portalgo_reserve6, 0, NULL, 0,
   1771 		       CTL_CREATE, CTL_EOL);
   1772 	sysctl_createv(clog, 0, NULL, NULL,
   1773 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1774 		       CTLTYPE_INT, "neighborgcthresh",
   1775 		       SYSCTL_DESCR("Maximum number of entries in neighbor"
   1776 			" cache"),
   1777 		       NULL, 1, &ip6_neighborgcthresh, 0,
   1778 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1779 		       CTL_CREATE, CTL_EOL);
   1780 	sysctl_createv(clog, 0, NULL, NULL,
   1781 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
   1782 		       CTLTYPE_INT, "maxdynroutes",
   1783 		       SYSCTL_DESCR("Maximum number of routes created via"
   1784 			   " redirect"),
   1785 		       NULL, 1, &ip6_maxdynroutes, 0,
   1786 		       CTL_NET, PF_INET6, IPPROTO_IPV6,
   1787 		       CTL_CREATE, CTL_EOL);
   1788 }
   1789 
   1790 void
   1791 ip6_statinc(u_int stat)
   1792 {
   1793 
   1794 	KASSERT(stat < IP6_NSTATS);
   1795 	IP6_STATINC(stat);
   1796 }
   1797